Maggie Zhang of NBC Universal invited all the presenters back to a wrap-up session called “Attribution Pivot,” where she asked what challenges marketers are facing and how they are meeting them. Each provided insight into important attribution challenges that they as a marketer or their client is facing. Limitations include lacking the ability to do A/B testing, privacy issues and the looming issue of cookie depreciation. It is also difficult to determine long-term lift, such as lifetime value.
Stu Schwartzapfel of iSpot.tv and Traci Will of Gamut talked about dealing with the challenges of local OTT attribution. While national brands have been a fixture on streaming platforms for a while now, smaller and local brands are just beginning to dip their toes into this space. Industry challenges include a lack of standardization of measurement, a national versus a local focus and careless measurement which can have confusing results.
Hong Zou of Adobe talked about how they measure campaign incrementality and ROI using both first-party and third-party identifiers. Their method was born in 2020, out of the need to refocus from lower funnel marketing campaign performance to upper funnel performance. Adobe matches one group of people exposed to the campaign with a look-alike group who were not exposed. Since all other attributes are the same, any differences can be attributed to the campaign itself.
Victoria Schiappacasse from Mercado Libre, a Latin American ecommerce technology company, described the reasoning behind and steps involved in implementing Meta’s open-source GeoLift, an incrementality application, in-house. As introduced by Meta’s Nicolas Cruces, the advantages of the free software bring value by “democratizing measurement.” Flexibility, user-friendliness and a robust support community are some of the other main advantages cited by Victoria and exemplified in two use cases.
With Trevor Testwuide (Measured) providing context and Ian Yung (Tonal) guiding the case studies from Pinterest and Google, the two presenters tested whether ads/channels were working and how far marketers can scale them. Trevor compared last-touch attribution to incremental ROAS, showing the significant discrepancies between platform-reported and media-driven incremental conversions. The incrementality experiment methodologies used in the case studies were cohort based first-party audience split testing and geo-matched market incrementality testing, which Trevor noted were must-haves in determining where to cut waste and where to scale. Results from the case studies measuring holdout cohorts showed overinvestment in Pinterest based on organic conversions, and a 13% increase in ROAS on Google Shopping from under-reporting of incrementality.
In providing an overview of retail media’s latest trends, opportunities and challenges for retailers and brands, Michael Ellgass (Circana) shared results from analyses of over 100 CPG studies that looked at channel performance in retail media networks and the halo impact outside those networks. Finding that the total incremental sales impact is often larger than what the retail media can see in their own outlets and that most shoppers are influenced by a variety of channels, Michael presented the nuanced data that supported a combined approach for maximum impact.
In this session, Tania Yuki and Brian Pugh of Comscore explored the impact of frequency and latency in cross-platform advertising effectiveness. In her opening, Tania demonstrated consumer trends and touchpoints to better understand cross-media, in terms of reach and optimizing platforms for specific outcomes. In her discussion, Tania acknowledged the challenges of measurement due to the constant introduction of new innovations and the adoption of new behaviors to track. She also recognized the considerable increase in connected devices per household since the pandemic. Tania pointed out complexities in the current media ecosystem from the increase in which media has merged despite being separate platforms (e.g., linear TV, social media, online video, etc.). In addition to all the changing behavior in media consumption, the speaker noted the emergence of Generation Z is beginning to change the rules for establishing brand love and loyalty. In his discussion, Brian examined findings from the measurement of 400 cross-platform campaigns to understand trends in terms of platform mixes. Brian noted the continued growth of social media and CTV along with the decline in linear TV, though he acknowledged linear still remained “king.” Furthermore, he found that multi-screen campaigns performed better than single-platform campaigns.
In this session, Yannis Pavlidis of consumer insights and CX firm DISQO tackled the challenges of benchmarking, cross-media outcomes and brand lift due to incomplete data from siloed platforms and media channels. In the opening, Yannis provided a refresher on the importance of benchmarks and obstacles from existing approaches to benchmarking (e.g., inconsistent methodologies, outdated data and collection techniques). The discussion examined solutions to address issues in data collection concerning benchmarking ad impact, which streamlines the process using consented, single-source data. The presentation also examined calculating benchmarks based on data taken from one source group (rather than two unaffiliated groups), considered the recency of the campaign used and subsequent behavior(s) which then can be correlated with survey responses. The advantage of using consented single-source data is that it can lead to more insightful, relevant and consistent outcomes in benchmarks.
In this session, speakers Bennett M. Kaufman, Kyle Holtzman and Michelle Smiley of Google explored a two-pronged approach to cross-media measurement and planning that considered the full-funnel impact across traditional TV and streaming video (YouTube), to make sense of all the “disparate forms of data and measurement.” The approach considered a geo-based experiment and audience incrementality to demonstrate and solve the following challenges: to retain current loyal customers, to age down the brand and to appeal to new consumers (Generation Z). The speakers presented a study done by Google in partnership with Burger King to test a new experimentation strategy to understand and measure the relationship between Linear TV and YouTube. The speakers touted the benefits of this method as repeatable and customizable across a variety of media channels, in addition to being timely, omni-channel and privacy safe.
Kyle Holtzman – Business Lead Restaurant Vertical, Google
Bennett M. Kaufman – Cross-Media Measurement Lead, Google/YouTube
Michelle Smiley – Analytical Lead Restaurant Vertical, Google
In this session, speakers Bennett M. Kaufman, Kyle Holtzman and Michelle Smiley of Google explored a two-pronged approach to cross-media measurement and planning that considered the full-funnel impact across traditional TV and streaming video (YouTube), to make sense of all the "disparate forms of data and measurement." The approach considered a geo-based experiment and audience incrementality to demonstrate and solve the following challenges: to retain current loyal customers, to age down the brand and to appeal to new consumers (Generation Z). The speakers presented a study done by Google in partnership with Burger King to test a new experimentation strategy to understand and measure the relationship between Linear TV and YouTube. The speakers touted the benefits of this method as repeatable and customizable across a variety of media channels, in addition to being timely, omni-channel and privacy safe.
Key Takeaways
The geo-based experiment addressed the understanding of changing behavior in the physical stores for Burger King, through increased sales related to media spend. This technique gave the ability to measure the uplift between control and treatment to understand media impact. The geo-experiment focused on three KPIs: store sales, store transactions and deal take rate (promotion featured in the ad).
Results from the geo-experiment indicated:
Store sales generated by linear TV were flat but store sales increased in views from YouTube.
Store transactions generated by linear TV decreased while YouTube views increased store transactions.
In terms of the deal take rate (deal shown in the ad) the take rate was higher generated by linear TV, though it still generated positive returns from YouTube.
Audience incrementality testing was conducted by Comscore (3rd party incrementality validation). Through this process, they wanted to understand if they were reaching a new target audience and if their message was reaching anyone that may not have heard their message on linear TV alone.
Audience incrementality testing resulted in the following:
Accounting for the target audience of adults 18-49 was critical in the short and long term.
YouTube reached 78 million adults ages 18-49. In addition, 34 million of the viewers were YouTube-only, unique viewers.
There were 43,365,489 cross-platform unique viewers.
20,680,526 were unique linear TV-only viewers with 64 million total linear TV viewers.
